93% of Capital City Suburbs Fell Through Winter. The Refinance That Dies Now Dies on the Valuation, Not the Calculator

The headline from Cotality’s August Home Value Index is that national values fell 0.9%, a fifth consecutive monthly decline. The number brokers should actually be working from is 93% — the share of capital city suburbs that recorded a fall through winter, up from 45.8% in autumn.

The number that matters is 93%, not 0.9%

Cotality released its Home Value Index for August on 1 September. Nationally, values fell 0.9% over the month and now sit 3.6% below the peak reached in March 2026. It was the fifth straight month of declines.

Sydney led the capitals down 1.4% and is now 7.1% below its February peak. Melbourne and Canberra each fell 1.1%, Brisbane 1.0%, and Adelaide and Perth 0.8%. Combined regional markets held up better, easing 0.4% in August and 1.2% across winter against 1.4% for the combined capitals.

Those are the figures that will be quoted in client emails this week. But the one that should change how you triage files is the breadth measure: through winter — the three months to 31 August — 93% of capital city suburbs recorded a decline in values. In autumn, that figure was 45.8%. In a single season, a downturn that was concentrated enough to manage suburb by suburb became close to universal.

The supply and demand numbers underneath explain why. Capital city advertised stock is running 24% higher than a year ago and 8% above the five-year average, while the flow of new listings is actually 6% lower year-on-year. In other words, the build-up is not a wave of fresh vendors — it is stock that is not clearing. Quarterly sales are tracking 15.5% below the same period last year and 11.5% under the five-year average, with Brisbane, Perth and Sydney each down more than 20%.

Cotality research director Tim Lawless told the ABC that “the combination of a sharp drop in demand and higher than average advertised stock levels are weighing more heavily on Australia’s largest housing market.” Ray White chief economist Nerida Conisbee put the same shift in physical terms in the ABC’s report: open home attendance has gone from around four people on average last year to around two.

Why breadth changes the broker conversation

Depth and breadth do different things to a loan book.

A deep fall in a small number of markets is a scenario problem. You know which postcodes are soft, you flag them, you order a valuation early on those files, and the rest of the book behaves normally. That is roughly how the last two years have worked for most residential brokers: a handful of off-the-plan unit exposures, some outer-ring house and land, a few resource-town anomalies, and everything else valuing at or above expectation.

A shallow fall across almost every market is a different animal. It is not severe enough to make headlines in any one suburb, which is precisely why clients will not have priced it in — but it removes the quiet assumption that has been sitting underneath a large share of residential files: the security will value up.

This matters because brokers rarely see the valuation until late. The client’s expectation of their property’s worth is usually set at the first appointment, often from a portal estimate or a neighbour’s sale price from 2024. Every subsequent step — the structure, the LVR, the lender shortlist, the cash-out figure — inherits that number. When it is wrong by five or seven per cent, it is not a small correction. It is frequently the difference between a 78% LVR and an 82% one, and everything that follows from crossing that line.

The constraint has moved from servicing to security

For much of the past two years, the conversation that killed a refinance was servicing. Capacity was the bottleneck, the assessment rate buffer was the villain, and the broker’s skill was in finding the lender whose policy treated the client’s income most generously. Security was assumed and rarely interrogated until the valuation came back.

The August data suggests a second constraint is now live on a much larger share of files, and it binds independently of the first. A client can pass servicing comfortably and still be unable to move, because the LVR the valuation produces puts the product out of reach — or drags lenders mortgage insurance into a transaction that was priced without it.

The arithmetic is unforgiving in a way that servicing is not. Servicing failures are usually gradual: you are short, you find a lender with a different treatment of overtime or rental income, and you get there. LVR failures are step functions. Most lenders price LMI in above 80% LVR, and pricing and policy tiers commonly step again in the high-80s and at 90%, though bands vary by lender. A property that has drifted down four or five per cent does not make the loan slightly harder. It can move the file across a band and change the product entirely.

Consider a client who purchased in Sydney early in 2026 with a 12% deposit. On the figures above, the market they bought into is now 7.1% below its February peak. Nothing about their income has changed and nothing about their conduct has changed — but the equity position that made a refinance sensible on paper in March may not be there in September. That is not a distressed borrower. That is an ordinary file, and there are a lot of them.

The five file types where this bites first

Not every file is exposed. The ones that are share a common feature: the transaction only works if the property is worth what the client thinks it is worth.

• Rate-driven refinances above 80% LVR. The most common casualty. The client is switching to save money, the saving is calculated without LMI, and a lower valuation reintroduces a premium that swallows the benefit. Worth noting that this cuts both ways — a client already paying LMI-affected pricing may have less to gain from moving than your comparison suggests.
• Cash-out and equity release. Renovation funding, a deposit for the next purchase, business capital. These are usually sized off an assumed value. When the value lands lower, the available cash-out shrinks by a multiple of the shortfall, because the client is drawing against the top slice of the LVR band.
• Debt consolidation. The consolidation only clears the debts if the security supports the new loan amount. A short valuation can leave the client with a partially consolidated position that is worse than where they started — and this is the category where the pressure to “make it work” is highest.
• Guarantor releases. A family pledge comes off when the LVR on the primary security drops below the lender’s threshold. If values are falling, that date is moving away from the client rather than towards them. If you have promised a release “in about a year”, that promise deserves revisiting.
• Upgraders and simultaneous settlements. Falling values compress equity at both ends, and the listings data says the sale side is slower. Quarterly sales are down 15.5% year-on-year, and stock is sitting. Timing assumptions built when properties were selling in weeks need re-testing.

Where best interests duty sits in this

None of this is a compliance story on its own. But it does intersect with the way ASIC frames a broker’s process.

ASIC’s Regulatory Guide 273, Mortgage brokers: Best interests duty, published in June 2020, sets out the regulator’s guidance on the duty. It says at RG 273.16 that brokers “must act in the best interests of each individual consumer to whom credit assistance is provided”, and at RG 273.44–45 that after gathering information a broker is “likely to need to consider the available products and assess which of those (if any) are in the consumer’s best interests”.

Two passages are worth re-reading in the current market. At RG 273.59–61, ASIC addresses situations where access to certain products may be limited — brokers should explain the limitation and still consider the cost of available options where approval is reasonably likely. And at RG 273.50: “If you cannot act in the consumer’s best interests in providing them with credit assistance, you must not provide the assistance.”

A security-first order of operations

The change most brokers need to make is small and sequencing-based. It costs nothing and it moves the disappointment from week three to day one, where it is survivable.

1. Ask what the client thinks it is worth — and where that number came from. A portal estimate, a 2024 appraisal and a neighbour’s sale are three different qualities of evidence. Write down which one you were given.
2. Test the number before you test capacity. Use whatever your aggregator’s platform offers for automated estimates, and pay attention to the confidence indicator rather than only the midpoint. Where a lender offers a free upfront valuation, order it before you shortlist, not after.
3. Model the LVR band, not the LVR. Work out what value the file needs to stay under 80%, and how far the valuation can fall before the structure changes. If that buffer is thin, say so at the first appointment.
4. Price the downside version. Prepare the comparison with LMI included as well as excluded where the client is near a threshold, so a lower valuation is a scenario you already discussed rather than bad news you deliver.
5. Know your panel’s valuation behaviour. Which lenders will accept a desktop or automated valuation and at what LVR, which will escalate to a full inspection, and which allow you to submit comparable sales in support. This is now a placement factor, not an operational detail.
6. Build the comparable-sales habit. Where a valuation lands short, evidence is what moves it — recent, genuinely comparable settled sales, not argument. Set the expectation with the client that a challenge is possible but not assured.

What to review this week

A focused back-book exercise will tell you the size of your exposure faster than any market commentary.

• Pull every file settled in the last 24 months at 85% LVR or above with security in Sydney, Melbourne, Canberra or Brisbane. These are the markets that fell hardest in August and the LVR band with the least room.
• Re-run current estimated values and recalculate the LVR on today’s balance. Sort by how far each file sits from the 80% line.
• Flag any client you have told, verbally or in writing, that they would be able to refinance, release a guarantor or draw equity “next year”. Those conversations need updating before the client makes plans around them.
• Check your pipeline for cash-out and consolidation files where the value assumption has not been tested. Test it now, while the structure can still change.
• Review how your fact-find records the source of a property value estimate. If it does not, add the field.

What to watch next

Three things will tell you whether this is a seasonal soft patch or a longer reset. The first is whether the listings overhang clears into spring: advertised stock is 24% above a year ago while new listings are 6% lower, so the next two months will show whether buyers absorb the backlog or vendors withdraw. The second is regional divergence — regional markets fell 0.4% in August against the capitals’ steeper falls, and if that gap holds it becomes a genuine placement consideration rather than a talking point. The third is lender valuation policy: if automated and desktop valuations start returning lower confidence scores at scale, panels will lean harder on full inspections, and turnaround times will move with them.

None of that is knowable today. What is knowable is the sequencing change, and it does not require a forecast to justify. In a market where nine capital city suburbs in ten went backwards over winter, the broker who establishes the security position before building the recommendation is simply working from the market that exists rather than the one their client remembers.

Sources: Cotality Home Value Index, August 2026 (released 1 September 2026), as reported by ABC News and Property Update; ABS Lending Indicators, June quarter 2026 (released 14 August 2026); ASIC Regulatory Guide 273, Mortgage brokers: Best interests duty (June 2020). Quotes from Tim Lawless (Cotality) and Nerida Conisbee (Ray White) as reported by ABC News.